Home
150 Teaching Data Enable Robot Adaptation: Ant Lingbo Open-Sources LingBot-VLA Fine-Tuning Code

Ant Group-owned robotics firm Lingbo Technology has announced the full open-source release of the real-robot post-training toolchain for its embodied foundation model, LingBot-VLA. This toolchain lets development teams quickly adapt LingBot-VLA to their own robots and tasks using their own data.
The growing number of open-source embodied intelligence models still requires extensive adaptation work to deploy on a specific robot. Differences in robotic arm configurations, end-effectors, sensor setups, and control interfaces mean teams often must perform significant engineering for real-robot deployment. Such workflows have typically been kept as internal know-how and rarely fully open-sourced before.
This release targets key real-robot adaptation needs across four stages: data processing tools that support merging multiple LeRobot datasets and standardize joint dimension mapping, training configurations optimized for real-robot scenarios, offline evaluation tools, and a real-robot deployment module with compilation acceleration. The model also offers versions with and without depth, letting teams choose based on their requirements.
As an embodied foundation model, LingBot-VLA is pre-trained on 20,000 hours of real-robot data covering nine mainstream dual-arm robot configurations. It possesses cross-body and cross-task generalization capabilities. In both real-robot and simulation evaluations, it outperforms the industry benchmark π0.5 and has undergone multi-machine verification with manufacturers such as Lepu, Songling, and Xinghai Tu.
LingBot-VLA can achieve high-quality task transfer with just 150 demonstration data points. Thanks to deep optimization of the underlying codebase, its training efficiency is 1.5 to 2.8 times that of mainstream frameworks like StarVLA and OpenPI, further reducing data and computing costs for model adaptation.
The LingBot-VLA code repository is now open-sourced on GitHub (github.com/Robbyant/lingbot-vla), and model weights are available on Hugging Face and ModelScope.
Related article
ChatGPT Deep Integration Unlocks 300,000 Movies and TV Shows via Conversation
As streaming services saturate the market with endless content, helping users cut through the noise to find what they actually want has emerged as a critical industry hurdle.On April 9, major streaming providers officially announced the integration o
How to fix Core Web Vitals for better SEO ranking
Transform your video editing workflow with AI-powered transcriptsTable of Contents:IntroductionKey Advantages of Automated TranscriptionEnsuring Accuracy and Refining OutputLeveraging Search Capabilities in TranscriptsSeamless Import of Existing File
DingTalk Rebuilds Architecture With AliAI’s MuleRun, Launching Full-Agent Era
DingTalk’s new CEO and head of the Wukong Business Unit, Chen Yusen, revealed a major organizational restructuring in an internal memo issued this afternoon. By creating the Core Platform Business Department and fully merging the AI Agent team, DingT
Related Special Topic Recommendations
Comments (0)
0/500

Ant Group-owned robotics firm Lingbo Technology has announced the full open-source release of the real-robot post-training toolchain for its embodied foundation model, LingBot-VLA. This toolchain lets development teams quickly adapt LingBot-VLA to their own robots and tasks using their own data.
The growing number of open-source embodied intelligence models still requires extensive adaptation work to deploy on a specific robot. Differences in robotic arm configurations, end-effectors, sensor setups, and control interfaces mean teams often must perform significant engineering for real-robot deployment. Such workflows have typically been kept as internal know-how and rarely fully open-sourced before.
This release targets key real-robot adaptation needs across four stages: data processing tools that support merging multiple LeRobot datasets and standardize joint dimension mapping, training configurations optimized for real-robot scenarios, offline evaluation tools, and a real-robot deployment module with compilation acceleration. The model also offers versions with and without depth, letting teams choose based on their requirements.
As an embodied foundation model, LingBot-VLA is pre-trained on 20,000 hours of real-robot data covering nine mainstream dual-arm robot configurations. It possesses cross-body and cross-task generalization capabilities. In both real-robot and simulation evaluations, it outperforms the industry benchmark π0.5 and has undergone multi-machine verification with manufacturers such as Lepu, Songling, and Xinghai Tu.
LingBot-VLA can achieve high-quality task transfer with just 150 demonstration data points. Thanks to deep optimization of the underlying codebase, its training efficiency is 1.5 to 2.8 times that of mainstream frameworks like StarVLA and OpenPI, further reducing data and computing costs for model adaptation.
The LingBot-VLA code repository is now open-sourced on GitHub (github.com/Robbyant/lingbot-vla), and model weights are available on Hugging Face and ModelScope.
ChatGPT Deep Integration Unlocks 300,000 Movies and TV Shows via Conversation
As streaming services saturate the market with endless content, helping users cut through the noise to find what they actually want has emerged as a critical industry hurdle.On April 9, major streaming providers officially announced the integration o
How to fix Core Web Vitals for better SEO ranking
Transform your video editing workflow with AI-powered transcriptsTable of Contents:IntroductionKey Advantages of Automated TranscriptionEnsuring Accuracy and Refining OutputLeveraging Search Capabilities in TranscriptsSeamless Import of Existing File
DingTalk Rebuilds Architecture With AliAI’s MuleRun, Launching Full-Agent Era
DingTalk’s new CEO and head of the Wukong Business Unit, Chen Yusen, revealed a major organizational restructuring in an internal memo issued this afternoon. By creating the Core Platform Business Department and fully merging the AI Agent team, DingT











